Amazon's Custom AI Chip Capacity Nearly Sold Out as AWS Secures Cerebras Deal
Amazon’s in-house AI chip computing capacity is reported to be nearly sold out, underscoring surging AWS demand for custom silicon. AWS also agreed to deploy Cerebras Systems’ wafer-scale AI accelerator in its data centers after Cerebras raised $1.1 billion at an $8.1 billion valuation and filed for a mid-May IPO.
1. Amazon Custom Chip Capacity
Amazon’s internally developed AI chips have reached nearly full capacity utilization, with AWS reporting that available computing nodes for its custom hardware are almost completely booked by enterprise and internal workloads. This level of adoption indicates strong customer demand for tailored silicon solutions to power machine-learning services.
2. AWS Partnership with Cerebras
AWS has secured an agreement to integrate Cerebras Systems’ wafer-scale AI accelerators into its data centers, expanding its hardware offerings for high-performance inference and training tasks. The deal follows Cerebras’ $1.1 billion Series G funding at an $8.1 billion valuation and precedes the startup’s planned mid-May IPO.
3. Competitive and Strategic Implications
The combination of near-sold-out internal chip capacity and the Cerebras deployment deal positions AWS to diversify its silicon stack and intensify competition with Nvidia. These moves could enhance AWS margins, attract more AI-centric customers, and strengthen Amazon’s strategic edge in cloud computing services.